Learning Forum

[ari003]

The relaxation factor is generally reduced to dampen the residuals or the monitor curve. But it has cons in doing so, it generally takes longer time to converge the solution or in the monitor curve(Cl) to generate a stable line. 

For example- The change in Cl curve(monitor curve) that is obtained in default relaxation factor is 0.1 in 10 iteration.

                       The change in Cl curve that is obtained after reducing the relaxation factor is 0.1 in 1000 iteration. So it seems like a straight                               line. In other word the change is gradual(Very less slope).

Then my doubt is how can the relaxation factor be of any help to the unstable solution? 

[Rob]

Because if the solution is unstable you won't get a result (the solver will diverge) with a reduced under-relaxation factor you'll get a solution, it just may take longer. 

[ari003]

When can I understand that it is my desired result? 

Sometimes it happens like the monitor curve remains unstable but the residuals seems to be converged, so in that case I cant consider it my desired result right?

Reducing the Courant number also stabilises the solution but I am unaware of its lower limit.

[DrAmine]

Imbalances are small and monitor profiles are not changing anymore for steady state runs.
Imbalances plus accumulation are zero (hard to verify) and monitor profiles not changing at the end of the time step.

For both residuals are nice to have that they get lower.